I can make my avatar male or female, accept the mission or reject it, negotiate with the bandits or kill them, use the shotgun or the katana, take my reward or demand more, can play it through on the good path, the evil path, use stealth or go in with guns blazing, play it alone or with others, play it on easy or on hard and on my platform of choice. Games like to be all things to everybody, appealing to the broadest possible demographic or perhaps a collective lack of attention.
Think about a game that refuses the player the above agencies/freedoms/liberties/inalienable rights and instead gives them only a knife then locks them in a basement with a monster. That narrow scenario demands our full attention in a way that less-linear games do not. We don't get to think about preferred weapons or the smartest strategies, our attention is dedicated to escaping with the very limited tools at our disposal. If the knife isn't working, we can't then fall back on the rocket launcher; we have to make it work. We have no alternatives, so we don't focus on how we like to play games, but on playing the game.
Survival horror can sustain that intensity because they don't ask you to make as many decisions. Games based around decision-making gameplay -- a tactical shooter or strategy game, or a role-playing morality test -- never evoke that same high-adrenaline sensation; the kind where your heart is pounding and you're completely fixated on your own survival since the consequences of failure are permanent. They're not capable of that but they pretend they are.
It's a cliche of marketing campaigns to promise that this time, in this game your choices really matter, and they affect the people and world around your character. You are given great latitude in how you build your defenses, prepare your troops and how you resolve a domestic dispute, and your decision will have lasting consequence. "Every choice matters," but have they, ever? In those genres where decisions should matter the most, they matter the least.
Drawing attention to multiple choices in single-player games reveals how compromised your decisions actually are. In a single-player game, your choices are never permanent, because you can always go back and try it again. Tough ethical quandaries and desperate battle tactics are easily redacted. Decisions are never final because replayability exists, even if the game doesn't have replay value in the conventional sense. Making a decision when prompted is more like reconnaissance than commitment. You can always take it back, and you know this, so you don't take it so seriously in the first place. You're aware of the alternatives and however subliminally, you register the point of no return as an opportunity to test your plan before loading a save and making your "final" choice. When Bioshock responds to my treatment of the Little Sisters, it's not judging the person I am. If I started killing Little Sisters, regretted it, reloaded, then it's judging the second-draft person I want the game to think I am.
The game tells me that the fate of the universe depends on my actions, but I don't appreciate that gravity, and my moral decisions are made with no conviction. If I don't like it, I'll change it. It's never a real choice if you know you have infinite chances. Choices never lead to last-ditch fights to the death where you will persevere even as the situation takes the slightest turn for the worse. Games construct that artifice but instead of overcoming the odds, you'll reload instantly.
Checkpoint saves are one way to enforce long-term consequence but they pay a price, mostly because the player is accustomed to the convenience of the quicksave. If the player screws up in a boss fight, and has to replay through a lengthy section of game, it'll diminish their interest rather than strengthening their resolve to meet the challenge.
It's very hard for single-player games to compel fight-or-flight responses in the player, and they don't need to, either. The problem is that it's their aspiration. Games are trending towards realism and simulation and every step forward towards a "living, breathing world" invokes the rhetoric about life-changing choices with real-world impact. Any progress made towards some nebulous "realism" goal is uncomfortably ignoring that certain elements of games are fundamentally unreal.
Quicksave abuse is something we assume to be an inherent video game limitation, but the rewind features of Braid and Prince of Persia: The Sands of Time co-opt that weakness and institute game impermanence as a feature. They acknowledge that games are intrinsically about infinite variations and constant revisions, and that there's this disparity between austere presentation and safety-net mechanics.
There have to be additional solutions. Bioshock thought it had one with its respawning chambers: it knew that players would reload anyway so it gave them a canonical reason not to fear death, but the internet hardcore yelled that they hated it. It removed the challenge, they said. It's 2008, let's stop pretending that challenge is still there. Finality is not permanent anymore. Difficult choices are meaningless. Instead of living with it, let's work with it, and decide now where games go next.